[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

RE: [DotGNU]RDF Query Language

From: Danny Ayers
Subject: RE: [DotGNU]RDF Query Language
Date: Wed, 26 Mar 2003 11:15:48 +0100

Hi Peter,

I'm only lurking (interested in the RDF developments) but I'm curious why
you suggest creating another query language, rather than reusing an existing
one - I believe the syntax of both RQL and RDQL is very similar to your


> -----Original Message-----
> From: Peter Minten [mailto:address@hidden
> Sent: 25 March 2003 18:22
> To: DotGNU Developers Mailing List
> Subject: [DotGNU]RDF Query Language
> Hi folks,
> for the GNU.RDF project ( an RDF Query
> Language (RQL)
> is needed. Here's what I have in mind. A query consists of
> roughly 2 parts: the
> actual query and metadata about the query. The query metadata can
> be used to
> notify the RDF server of things, like that only the local store of the RDF
> server should be searched and not the complete semantic web. RQL
> resources have
> one property query that contains the complete text of the query, all other
> properties are metadata. The output of an RDF query is the wanted RDF.
> The query text looks like SQL. A RDF resource with properties can
> be thought of
> as a tuple (table row), the columns being the properties. A
> property bag (a
> property that has as value a bag) can be thought of as a table
> with one column
> (the property) and many rows, the cells containing the value of
> the elements of
> the bag. All RDF containers and collections are represented this
> way. Note that
> in the resource type bag the order of the tuples is unspecified
> (because it's
> irrelevant). A RDF resource with beside other properties a sequence can be
> thought of as a table with on each row the same values for the
> non-container/collection properties and another value for the
> sequence property.
> Logical consequence of this is that when there a resource has two bags as
> property values the table can get quite big (sizeof(bag1)*sizeof(bag2)).
> Now to some syntax. The first difference from SQL is ALIAS, which is a
> pre-processor directive that works just like in C, it allows
> using simple names
> instead of uri's, note that ALIAS does not bluntly replace every
> occurence of a
> string, it only replaces if the string completely matches. Before
> sending the
> RQL to the server the client pre-processes.
> The basic SELECT operation goes like this:
> SELECT propertynames
> FROM resourceclassnames
> WHERE condition
> GROUP BY propertynames
> HAVING condition
> An example:
> ALIAS dc;
> ALIAS rdf;
> SELECT dc:creator
> FROM rdf:Description
> WHERE dc:language = "english";
> This query returns all the resources from class rdf:Description
> ( that
> have as language
> English and it returns the creator of those descriptions.
> That's the general idea, any comments before I develop this further?
> Greetings,
> Peter
> _______________________________________________
> Developers mailing list
> address@hidden

reply via email to

[Prev in Thread] Current Thread [Next in Thread]